The present invention relates to a process for preparing phthalic anhydride in which the catalytic gas-phase oxidation of o-xylene and/or naphthalene is carried out over at least three zones of catalysts of increasing activity and in which the increase in activity of the zones is effected in a particular way.
It is known that many carboxylic acids and/or carboxylic anhydrides are prepared industrially by the catalytic gas-phase oxidation of aromatic hydrocarbons such as benzene, the xylenes, naphthalene, toluene and durene in fixed-bed reactors, preferably multitube reactors. This is used to obtain, for example, benzoic acid, maleic anhydride, phthalic anhydride, isophthalic acid, terephthalic acid or pyromellitic anhydride. This is generally carried out by passing a mixture of a gas comprising molecular oxygen, for example air, and the starting material to be oxidized through a plurality of tubes arranged in a reactor. Each of the tubes contains a bed of at least one catalyst. To regulate the temperature, the tubes are surrounded by a heat transfer medium, for example a salt melt. Despite this thermostatting, hot spots in which the temperature is higher than in the remainder of the catalyst bed may occur. These hot spots cause secondary reactions such as total combustion of the starting material or lead to formation of undesirable by-products which can be separated from the reaction product only with difficulty, if at all, for example the formation of phthalide or benzoic acid in the preparation of phtahlic anhydride (PA) from o-xylene. Furthermore, the formation of a pronounced hot spot prevents rapid running-up of the reactor, since the catalyst can be irreversibly damaged above a certain hot spot temperature, so that the loading can only be increased in small steps and the increase has to be monitored very carefully (hereinafter referred to as running-up phase).
To decrease the intensity of these hot spots, it has become customary in industry to arrange catalysts having different activities in zones in the catalyst bed, with the less active catalyst generally being located in the fixed bed in a position where the reaction gas mixture comes into contact with it first, i.e. it is at the gas inlet end of the bed, while the more active catalyst is located toward the end where the gas leaves the catalyst bed (DE A 2546268, EP 286 448, DE 2948163, EP 163 231). The catalysts of differing activity in the catalyst bed can be exposed to the reaction gas at the same temperature, but the two zones comprising catalysts of differing activity can also be thermostatted to different reaction temperatures for contact with the reaction gas (DE A 2830765). According to EP 163 231, a plurality of the measures mentioned can be employed simultaneously for setting the activity structure described. It is known from WO 98/00778 that the addition of temporary activity dampers can lead to a shortening of the running-up phase. In EP 676 400, multistructuring in the reaction of tetraalkylbenzenes to give pyromellitic anhydride has been found to have a positive effect in respect of yield and product purity if the activity structuring is carried out such that the catalyst activity first increases and then decreases again in the flow direction of the gas. Finally, EP 99 431 states that, in the reaction of butane to give maleic anhydride, activity structuring of the catalyst bed has a positive effect on the productivity if the catalyst activity increases stepwise (or ideally continuously) in the flow direction of the gas; the activity structuring can be achieved by many different methods, preferably by dilution with inert material. As a result of the stepwise activity increase, a more homogeneous distribution of the energy liberated by the exothermic reaction can be achieved, so that larger amounts of MA can be produced. Since the reaction is carried out at partial conversion, the activity structuring can be achieved in virtually any way. These teachings cannot be applied to the preparation of PA by oxidation of o-xylene and/or naphthalene since, as is known, phthalic anhydride of good quality is obtained only when the reaction is carried out at full conversion (i.e.  greater than 99.9% conversion, based on the starting material used) in order to minimize contamination by undesirable, color-producing components such as phthalide or naphthoquinone and to avoid contamination of the waste gas by residual xylene or residual naphthalene.
EP-A 539 878 discloses a method for production of phthalic anhydride from o-xylene and naphthalene over a two-stage catalyst. The catalysts in both stages are so constructed that an inactive carrier is loaded with vanadium pentoxide, anatase titanium dioxide and also niobium, phosphorus, antimony and at least one element oxide of potassium, rubidium, cesium and thallium, the second-stage catalyst differing from the first-stage catalyst in containing less of said at least one element oxide of potassium, rubidium, cesium and thallium.
Despite the abovementioned proposals for improvement, long running-up times of 2-8 weeks or more have hitherto been necessary. xe2x80x9cRunning-up timexe2x80x9d describes the time which is needed to bring the catalyst to the desired final loading, according to the present state of the art 80 g of o-xylene/standard m3 of air or more, i.e. to bring the oxidation to the steady state, without the catalyst being damaged irreversibly. Here, particular care has to be taken to ensure that the hot spot does not exceed a certain critical value (usually 450-480xc2x0 C.), since otherwise the PA selectivity, the PA product quality and the life of the catalyst are very adversely affected.
It is an object of the present invention to find a process for preparing phthalic anhydride in which it is possible to simultaneously achieve all desired parameters such as short running-up time, high yield and low formation of by-products and also good product quality, e.g. a low phthalide content, by a combination of particular catalyst zones.
We have found that this object is achieved by a process for preparing phthalic anhydride by catalytic gas-phase oxidation of xylene and/or naphthalene by a gas comprising molecular oxygen in a fixed bed at elevated temperature and using at least three coated catalysts arranged in superposed zones, which catalysts have a layer of catalytically active metal oxides applied in the form of a shell to a core of support material, in which process the catalyst activity rises from zone to zone from the gas inlet end to the gas outlet end and the activity of the catalysts of the individual zones is set such that the least active catalyst comprises a lower amount of active composition and, if desired, additionally more alkali metal selected from the group consisting of potassium, rubidium and cesium as dopant than the catalyst of the next zone and the subsequent even more active catalyst comprises the same amount of active composition and even less alkali metal as dopant or a greater amount of active composition and, if desired, less alkali metal as dopant than the catalyst of the second zone, with the proviso that
a) the least active catalyst on nonporous support material comprises from 5 to 9% by weight, based on the total catalyst, of active composition comprising from 3 to 8% by weight of V2O5, from 0 to 3.5% by weight of Sb2O3, from 0 to 0.3% by weight of P, from 0.1 to 0.5% by weight of alkali metal (calculated as metal) and as balance TiO2 in anatase form having a BET surface area (J. Amer. Chem. Soc. 60 (1938), 309 et seq.) of from 18 to 22 m2/g,
b) the next more active catalyst has the same composition as catalyst (a) except for an active composition content which is from 1 to 5% by weight (absolute) higher and an alkali metal content which is from 0 to 0.25% by weight (absolute) lower and
c) the most active catalyst has the same composition as (a) except for an active composition content which is from 1 to 5% by weight (absolute) higher than in (a) and an alkali metal content which is from 0.15 to 0.4% by weight (absolute) lower than in (a).
Accordingly, the catalyst (a) newest to the gas inlet end always has a lower content of active composition (and possibly also an addition of alkali metal, in particular cesium, which decreases the activity) than the next catalyst (b) which in turn can have the same or a lower content of active composition than the subsequent catalyst (c). If the catalysts (b) and (c) have the same content of active composition, the alkali metal content of catalyst (b) has to be greater than that of catalyst (c). Subject to these conditions, the catalyst (a) has, according to a preferred embodiment, an active composition content of from 6 to 8% by weight comprising from 4 to 8% by weight of V2O5 and from 0.3 to 0.5% by weight of cs (calculated as Cs), in each case based on the active composition, the catalyst (b) has an active composition content of from 8 to 12% by weight comprising from 0.2 to 0.5% by weight of Cs, based on the active composition, and the catalyst (c) has an active composition content of from 8 to 12% by weight comprising from 0 to 0.3% by weight of Cs, based on the active composition. According to a particularly preferred embodiment, the catalyst (a) has an active composition content of from 7 to 8% by weight comprising from 6 to 8% by weight of V2O5 and from 0.3 to 0.4% by weight of Cs, in each case based on the active composition, the catalyst (b) has an active composition content of from 9 to 11% by weight comprising from 0.2 to 0.4% by weight of Cs, based on the active composition, and the catalyst (c) has an active composition content of from 9 to 11% by weight comprising from 0.05 to 0.2% by weight of Cs (in each case calculated as Cs), based on the active composition.
In place of mutually delineated zones of the different catalysts, it is also possible to achieve a pseudocontinuous transition between the zones and a pseudocontinuous increase in the activity by having a region in which one catalyst is mixed with the next at the transition from one zone to the next zone.
The catalysts (a), (b) and (c) used for the various zones are known per se and their preparation and composition has been described many times. In brief, these catalysts are generally coated catalysts in which the catalytically active composition is usually applied in the form of a shell to a nonporous support material which is generally inert under the reaction conditions, e.g. quartz (SiO2), porcelain, magnesium oxide, tin dioxide, silicon carbide, rutile, alumina (Al2O3), aluminum silicate, magnesium silicate (steatite), zirconium silicate or cerium silicate or mixtures of these support materials. The catalytically active constituent of the catalytically active composition of these coated catalysts is generally titanium dioxide in the form of its anatase modification together with vanadium pentoxide. In addition, the catalytically active composition may further comprise small amounts of many other oxidic compounds which act as promoters to influence the activity and selectivity of the catalyst, for example by lowering or increasing its activity. Examples of such promoters are the alkali metal oxides, in particular lithium, potassium rubidium and cesium oxide, thallium(I) oxide, aluminum oxide, zirconium oxide, iron oxide, nickel oxide, cobalt oxide, manganese oxide, tin oxide, silver oxide, copper oxide, chromium oxide, molybdenum oxide, tungsten oxide, iridium oxide, tantalum oxide, niobium oxide, arsenic oxide, antimony oxide, cerium oxide and phosphorus pentoxide. Promoters which reduce the activity and increase the selectivity are, for example, the alkali metal oxides, while oxidic phosphorus compounds, in particular phosphorus pentoxide, increase the activity of the catalyst but reduce its selectivity.
Such coated catalysts are produced according to the processes of DE A 1642938 and DE A 1769998 by spraying an aqueous and/or organic solvent-containing solution or suspension of the constituents of the active composition and/or their precursor compounds, hereinafter referred to as the xe2x80x9cslurryxe2x80x9d, onto the support material in a heated coating drum at elevated temperature until the desired proportion by weight of active composition in the total catalyst has been achieved. According to DE 2106796, coating can also be carried out in fluidized-bed coaters as are mentioned, for example, in DE 1280756. However, spraying in the coating drum or coating in a fluidized bed results in high losses since considerable amounts of the slurry are converted into a mist or part of the active composition already applied is rubbed off again by abrasion and are carried out by the waste gas. Since the proportion of active composition in the total catalyst should generally deviate only slightly from the specified value because the amount of active composition applied and thus the thickness of the shell strongly influence the activity and selectivity of the catalyst, in the production methods indicated the catalyst has to be cooled frequently, taken from the coating drum or the fluidized bed and weighed to determine the amount of active composition applied. If too much active composition is applied to the catalyst support, it is generally not possible for the excess active composition to be carefully removed afterwards without adversely affecting the strength of the shell, in particular without crack formation in the catalyst shell.
To reduce these problems, it has become customary in industry to add organic binders, preferably copolymers, advantageously in the form of an aqueous dispersion, of vinyl acetate/vinyl laurate, vinyl acetate/acrylate, styrene/acrylate, vinyl acetate/maleate or vinyl acetate/ethylene, to the slurry. The amount of binder used is 10-20% by weight, based on the solids content of the slurry (EP 744214). If the slurry is applied to the support without the addition of organic binders, coating temperatures above 150xc2x0 C. are advantageous. When the above-described binders are added, the usable coating temperatures are, depending on the binder used, from 50 to 450xc2x0 C. (DE 2106796). The binders applied burn off within a short time after introduction of the catalyst into the reactor and start-up of the reactor. Furthermore, the addition of binder has the advantage that the active composition adheres well to the support, so that transport and charging of these catalysts are made easier.
Further suitable methods of producing coated catalysts for the catalytic gas-phase oxidation of aromatic hydrocarbons to give carboxylic acids and/or carboxylic anhydrides are described in WO 98-00778 and EP A 714700. The layer comprising the catalytically active metal oxides is applied to a support material by first preparing a powder from a solution and/or a suspension of the catalytically active metal oxides and/or their precursor compounds, if desired in the presence of auxiliaries for catalyst production, subsequently applying the powder in the form of a shell to the catalyst support, if desired after conditioning and, if desired, after heat treatment for generating the catalytically active metal oxides, and subjecting the support which has been coated in this way to a heat treatment for generating the catalytically active metal oxides or a treatment to remove volatile constituents.
The conditions of the process for preparing phthalic anhydride from o-xylene and/or naphthalene are likewise known from the literature. In particular, reference may be made to a review in K. Towae, W. Enke, R. Jxc3xa4ckh, N. Bhargana xe2x80x9cPhthalic Acid and Derivativesxe2x80x9d in Ullmann""s Encyclopedia of Industrial Chemistry Vol. A. 20, 1992, 181, which is hereby incorporated by reference. Unlike the process known from the literature reference, the present process enables the xe2x80x9crunning-up timesxe2x80x9d, i.e. the times required to reach the steady state, to be reduced to, as a rule, from 5 to 20 days. Otherwise, the boundary conditions known from this literature reference and also, for example, from WO-A 98/37967 apply to the steady operating state of the oxidation.
For this purpose, the catalysts are first introduced into the reaction tubes of the reactor which are thermostatted from the outside to the reaction temperature, for example by means of salt melts. The reaction gas is passed over the catalyst bed prepared in this way at generally from 300 to 450xc2x0 C., preferably from 320 to 420xc2x0 C. and particularly preferably from 340 to 400xc2x0 C., at a gauge pressure of generally from 0.1 to 2.5 bar, preferably from 0.3 to 1.5 bar and at a space velocity of generally from 750 to 5000 hxe2x88x921.
The reaction gas passed over the catalyst is generally produced by mixing a gas comprising molecular oxygen, which gas may further comprise suitable reaction moderators and/or diluents such as steam, carbon dioxide and/or nitrogen, with the aromatic hydrocarbon to be oxidized. The gas comprising the molecular oxygen generally comprises from 1 to 100 mol %, preferably from 2 to 50 mol % and particularly preferably from 10 to 30 mol %, of oxygen, from 0 to 30 mol %, preferably from 0 to 10 mol %, of water vapor and from 0 to 50 mol %, preferably from 0 to 1 mol %, of carbon dioxide, balance nitrogen. To produce the reaction gas, the gas comprising the molecular oxygen is generally mixed with from 30 to 150 g of the aromatic hydrocarbon to be oxidized per standard m3 of gas.